O. Esanov

DEFENSE-IN-DEPTH FOR LARGE LANGUAGE MODELS: A MULTI-LAYER HYBRID ARCHITECTURE FOR PROMPT INJECTION DETECTION

A layered hybrid defense gateway that combines three complementary detectors: a fast prefilter based on lexical rules, a semantic classifier that uses word-level TF-IDF features together with a multilayer perceptron, and a character-level classifier that is robust to obfuscation is proposed.